6.1.3.4. Monitor Resources by Using Azure Monitor Insights
6.1.3.4. Monitor Resources by Using Azure Monitor Insights
💡 First Principle: Insights are curated monitoring experiences built on data Azure Monitor already understands. Some need nothing turned on, while VM insights needs an agent because the host can't see inside the guest operating system.
Scenario: Your manager wants one place to check whether VMs are running out of memory, which storage accounts are throttling requests, and whether the ExpressRoute connections are healthy, without writing KQL for each question.
What It Is: Azure Monitor Insights are prebuilt, workbook-based views for specific resource types, found under the Insights section of Azure Monitor (and on many resources' Monitoring menus). For AZ-104, focus on three.
VM insights (enhanced monitoring for VMs):
- Host metrics (CPU, network, disk I/O) are collected automatically at no cost.
- Guest metrics (memory, logical disks, processes) need enhanced monitoring. From the VM's Monitor page, select Configure. This installs the Azure Monitor Agent, assigns a system-assigned managed identity if needed, and associates a data collection rule (DCR).
- Choose a metrics experience: OpenTelemetry metrics (recommended; stored in an Azure Monitor workspace; a standard counter set at no cost) or logs-based metrics (classic; stored in the InsightsMetrics table of a Log Analytics workspace).
- Works for Azure VMs, scale sets, and Arc-enabled servers (Arc-enabled servers need the Azure Connected Machine agent first). Scale sets support guest monitoring through the Azure Monitor Agent but not the newer OpenTelemetry metrics experience. The Map feature and its Dependency agent are deprecated and retire June 30, 2028; performance monitoring isn't affected.
Storage insights:
- Shows availability, latency (E2E and server), transactions/errors, and capacity. It covers up to 200 storage accounts at a time from Azure Monitor > Insights > Storage Accounts, or one account from its Monitoring > Insights menu.
- Nothing to enable: it's built on storage metrics that are collected by default.
- Drill into the Failures, Performance, Availability, and Capacity workbooks. For example, look for
ClientThrottlingErrorin the failure breakdown.
Network insights:
- Shows topology, health and metrics for all deployed network resources, and alerts, with no configuration.
- The Connectivity tab shows your Connection monitor tests, the Traffic tab shows flow log and traffic analytics configuration, and the Diagnostic Toolkit opens Network Watcher tools such as IP flow verify, next hop, and packet capture.
| Insight | Needs setup? | Data source | Typical question |
|---|---|---|---|
| VM insights | Yes: Azure Monitor Agent + DCR | Azure Monitor workspace or Log Analytics | "Is this VM short on memory?" |
| Storage insights | No | Default storage metrics | "Which accounts are throttling?" |
| Network insights | No (tabs surface tests and logs you configure) | Resource health + platform metrics | "Which gateways are unhealthy?" |
⚠️ Exam Trap: If CPU charts work but memory charts are empty, the fix isn't a diagnostic setting. Platform metrics come from the host, and guest memory needs the Azure Monitor Agent through enhanced monitoring.
Reflection Question: Why can Storage insights and Network insights work with zero configuration while VM insights can't, and what does that tell you about host-level versus guest-level telemetry?
